#This code produces Table 3 and Figures 8 and 9 in appendix C
if(!dir.exists("figs")){dir.create("figs")}
if(!dir.exists("tabs")){dir.create("tabs")}

dir.create("tabs/k25")
dir.create("figs/k25")

library(tidyverse)
library(arabicStemR)
library(tidytext)
library(lubridate)
library(ggplot2)
library(quanteda)
library(stm)
library(stats)
library(ggthemes)
library(ggpubr)

rm(list = ls())
gc()

# 25 topics -------------------------------------------------------------
load("k25.RData")

meta = meta %>% mutate(document = article) %>% select(- article)

beta = tidy(topic_model)
gamma = tidy(topic_model, matrix = "gamma")

gamma = left_join(gamma, meta, by = "document")
gc()

terms_final = tibble(topic = character(),
                         terms = character())

terms_final = terms_final %>% 
  add_row(topic = "Media", terms = c("media, culture, public, union, first")) %>% 
  add_row(topic = "Assad and nation", terms = c("nation, president, people, army, Assad, martyrs")) %>% 
  add_row(topic = "ISIS", terms = c("terrorism, organization, ISIS, group, armed")) %>% 
  add_row(topic = "temperature and weather", terms = c("region, degree, temperature, south, north, governorate")) %>% 
  add_row(topic = "Israel and Palestine", terms = c("Israel, Palestine, occupation, Jerusalem, land")) %>% 
  add_row(topic = "terrorism", terms = c("terrorism, terror, army, armed, group, destroy")) %>% 
  add_row(topic = "europe", terms = c("France, Europe, Britain, Tunisia, first")) %>% 
  add_row(topic = "Russia and Iran", terms = c("Syria, Russia(n), states, foreign, Iran, Iranian")) %>% 
  add_row(topic = "Education", terms = c("university, education, public, governorate, students, study")) %>% 
  add_row(topic = "conspiracies and plots", terms = c("Syria, people, terrorism, Zionist, resistance, conspiracy")) %>% 
  add_row(topic = "legislation", terms = c("article, law, declaration, council, number, legislation")) %>% 
  add_row(topic = "bureaucracy", terms = c("council, public, project, minister/ministry, committee")) %>% 
  add_row(topic = "economy and development", terms = c("council, public, economy, project, sector, investment")) %>% 
  add_row(topic = "speeches", terms = c("said, president, states, discussion, politics")) %>% 
  add_row(topic = "Intl community and victims", terms = c("nations, united, humanitarian, organization, aid, rights")) %>% 
  add_row(topic = "Turkey", terms = c("Turkey, Erdogan, government, party, people, justice")) %>% 
  add_row(topic = "Lebanon", terms = c("Lebanon, resistance, Israel, army")) %>% 
  add_row(topic = "United States", terms = c("America, united, states, military, Washington, Obama")) %>% 
  add_row(topic = "Iraq", terms = c("Iraq, Baghdad, America, kill, injure")) %>% 
  add_row(topic = "Yemen and terrorism", terms = c("army, organization, military, terror, ISIS, Yemen,")) %>% 
  add_row(topic = "Gulf", terms = c("Saud, Qatar, politics, Gulf, world")) %>% 
  add_row(topic = "attacks", terms = c("armed, group, security, citizen, terrorist")) %>% 
  add_row(topic = "party politics?", terms = c("nation, work, economy, meeting, politics, citizen, party")) %>% 
  add_row(topic = "diplomacy", terms = c("president, relations, mister, visit, meeting, foreign")) %>% 
  add_row(topic = "religion", terms = c("Islam, religion, Patriarch, Muslim, Christian, Sheikh"))


topic_avg = gamma %>% 
  group_by(topic) %>% 
  summarise(avg = mean(gamma)) %>% 
  arrange(-avg)


terms_final = terms_final %>% mutate(topic2 = 1:nrow(terms_final)) %>% left_join(topic_avg, by = c("topic2" = "topic")) %>% select(-topic2)

names(terms_final) = c("Topic Label", "High Probability Terms", "Expected Proportion")

terms_final %>% arrange(desc(`Expected Proportion`)) %>% 
  xtable::xtable(caption = "Topic labels, highest probability terms, and expected proportion for n = 25 topics",
                 label = "tab:topics") %>% 
  xtable::print.xtable(include.rownames = F, size = "footnotesize",
                 file = "tabs/k25/topics.tex")


gc()

topics = tibble(topic = 1:25, `Topic Label` = terms_final$`Topic Label`)

gamma = gamma %>% ungroup %>% left_join(topics)

gamma = gamma %>% select(-topic) %>% rename(topic = `Topic Label`)

gamma = gamma %>% 
  select(document, topic, date, gamma) %>% 
  filter(nchar(topic) > 2) %>% 
  group_by(topic, date) %>% 
  summarise(avg = mean(gamma))


gamma2 = gamma %>% 
  spread(topic, avg)

gamma3 = gamma2 %>% 
  mutate(week = floor_date(date, "week")) %>% 
  select(date, week, everything()) %>% 
  group_by(week) %>% 
  mutate_if(.predicate = is.numeric, .funs = mean)


names(gamma3)[3:ncol(gamma3)] = paste0(names(gamma3)[3:ncol(gamma3)], "_fit")

gamma3 = gamma3 %>% left_join(gamma2)

plot_fun = function(y1){
  y2 = paste0(y1, "_fit")
  title = tools::toTitleCase(y1)
  
  df = gamma3[, c("date", y1, y2)] %>% as.data.frame()
  
  p = ggplot() + 
    geom_line(aes(x = df$date, y = df[,y1], colour = y1)) + 
    geom_line(aes(x = df$date, y = df[,y2], colour = y2)) + 
    geom_vline(xintercept = dmy("15-03-2011"), linetype = 2) + 
    scale_colour_manual(values = c("darkgrey", "black")) + 
    labs(title = title, y = expression(paste("Average  ", gamma)), x = "Year") + theme_few() + 
    theme(legend.position = "none", title = element_text(size = 20),
          axis.text = element_text(size = 14)) + 
    coord_cartesian(ylim = c(0, 0.25)) + 
    scale_x_date(date_breaks = "2 year", date_labels = "%y") + 
    scale_y_continuous(breaks = c(0, 0.10, 0.20))
  p %>% ggsave(filename = paste0("figs/k25/", y1, ".pdf"), width = 6, height = 5)
  return(plot)
}



plot_names = names(gamma3)[!str_detect(names(gamma3), "fit|date|week")]

for(i in 1:length(plot_names)){
  plot_fun(y1 = plot_names[i])
}
